(Carolyn Kaster / Associated Press) - President Obama pauses calls volunteers at field office for Organizing for America with Alexa Kissinger, left, and Suzanne Stern, right, in  Williamsburg, Va.

"It all comes down to the ground game."

That's perhaps the biggest cliche in electoral politics. And with national polls deadlocked and the election coming down to a few key swing states, it's more pertinent than ever. It also represents a key advantage for the Obama campaign, which has gotten many more supporters out to vote early than the Romney campaign has. But how much can a good get-out-the-vote (GOTV) effort really get a candidate?

The best work on this has been done by Yale's Alan Gerber and Columbia's Donald Green. They have run a series of randomized experiments to test the effectiveness of different GOTV techniques. They found that a lot of campaign mainstays are basically useless. E-mail appeals and robocalls, for instance, do nothing. Some mass media techniques, like nonpartisan radio, TV and newspaper ads, are effective and cost-efficient, but relatively little has been done in that area.

What works, they found, are personalized appeals. In-person canvassing is particularly effective, while professional and volunteer phone-banking could be the same (though there's less research on that). So campaigns that invest heavily in field operations, like President Obama's, are at an advantage. Seth Masket did the math and found that Obama's field operation swamps Romney's, as the below chart shows.

What's more, a Washington Post poll from earlier this month found that voters were likelier to be contacted by an Obama campaign staffer or volunteer than by a Romney staffer or volunteer, 23 percent to 20 percent, and 34 to 32 percent in swing states.

Political scientists generally think that GOTV matters less in presidential elections -- because the campaigns are likelier to be evenly matched -- than in lower key races for, say, dog catcher. But even so they've found some effects. Green and Joel Middleton noted that turnout among voters contacted by Obama's MoveOn.org in 2004 was 9 percent higher than among other voters. David Nickerson and Todd Rogers found that while traditional GOTV calls encouraging voting didn't work in the 2008 Pennsylvania primary, calls asking if respondents intended to vote did work, and calls asking where and when they intended to vote worked extremely well, increasing turnout by 9.1 percentage points. 

Perhaps the study that's most pertinent to the current election is Masket's paper on whether Obama's greater number of field offices helped him win the presidency in the 2008 race. Masket found that having an Obama office in a county was associated with a 0.8 percent boost in his share of the two-party vote in that county. McCain offices, by contrast, didn't have a statistically significant effect on the vote. And field offices were most effective in states that weren't saturated with them. Indiana, which had a minimal field presence, saw the biggest effects, while New Mexico, where most counties had field offices, saw smaller ones:

Source: Seth Masket

Now, Masket doesn't control for standing in the polls, so this doesn't suggest that Obama is going to outperform the polls by 0.8 points in counties where he has an office. But then again, polls don't capture the effects of a strong field team. "What the field offices do, or at least what they try to do, is turn an unlikely voter into a likely voter," Masket explains. "When pollsters are calling and the first thing they do is ask are you likely to vote, the people who've been contacted might just say no. If the field office is successful, that's where there might be a difference."

Masket's research suggests that McCain's field operation was unusually incompetent and that Obama's was unusually effective. Indeed, Masket says he has "a hard time imagining a much more effective organization than Obama '08." But then again, McCain was running a doomed campaign in a year when Republicans weren't particularly energized. The GOP is much more excited this time around, and has a much better shot at edging out the president, which means Romney might be getting more volunteers, and thus a stronger ground game, than McCain did four years ago.

"What really makes a field office successful is volunteers," Masket says, and Romney is better positioned now than McCain was.

Noting how hampered McCain was in 2008, and how impressive the Obama team was in using its field offices, Masket says that 0.8 points is about the best a ground team can hope to win. But in an election this close, that could make all the difference.